You are on page 1of 13

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/he

Effect of microstructure and composition on hydrogen


permeation in X70 pipeline steels

Ayesha J. Haq*, K. Muzaka, D.P. Dunne, A. Calka, E.V. Pereloma


School of Mechanical, Materials & Mechatronic Engineering, University of Wollongong, Wollongong, NSW 2522, Australia

article info abstract

Article history: The influence of microstructure and composition on permeation of hydrogen in 1.2 and
Received 21 June 2012 0.5 wt.% Mn X70 pipeline steels after different processing was investigated using an elec-
Received in revised form trochemical permeation technique.
9 November 2012 For 1.2 wt.% Mn (standard Mn) steel, the microstructure of normalised transfer bar was
Accepted 23 November 2012 coarse equiaxed ferrite grains. This sample exhibited the highest diffusivity, followed by
Available online 2 January 2013 transfer bar, with a mixed ferriteebainitic ferrite microstructure; and hot rolled strip, with
fine elongated ferrite grains.
Keywords: The 0.5 wt.% Mn (medium Mn) strip displayed lower diffusivity than the 1.2 wt.% Mn
Pipeline steel strip, due to hydrogen trapping by finer ferrite grains and a higher density of carbonitride
Permeability precipitates. Moreover, the medium Mn strip exhibited a uniform microstructure and
Diffusion consequently similar diffusion coefficients for the edge and centreline regions, whereas the
Traps finer grains of the edge region of the standard Mn strip resulted in a lower diffusivity than
Microstructure the centreline region.
Grain size Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.

1. Introduction hydrogen transport is an important step in the process of


embrittlement. Therefore, it is important to determine the
X70 pipeline steels are widely used to transport petroleum and diffusion rate of hydrogen in pipeline steels in order to predict
natural gas. The service life of pipeline steels is affected not their mechanical properties and in-service behaviour.
only by the fluid or gas which they transfer, but also by the The diffusion of hydrogen in steels is affected by the
environmental conditions under which they operate. One of microstructure of the steel: the phase or phases present, grain
the main reasons for the degradation in mechanical proper- boundaries, grain shapes, vacancies and dislocations, inter-
ties of steels is the exposure to hydrogen. For example, on faces with non-metallic inclusions, precipitate particles and
exposure to a sour gas environment, the steel surface corrodes voids [1e3]. These features can reduce the mobility of
and produces hydrogen at the surface. This hydrogen can be hydrogen by acting as traps. Traps are generally classified as
absorbed in the steel and can then start diffusing to regions of reversible and irreversible, based on the strength of their bond
stress concentration, impairing ductility and promoting brittle with hydrogen atoms. Grain boundaries and dislocations and
behaviour (hydrogen embrittlement). Since the transport of microvoids that have low trap binding energies are considered
hydrogen to a critical region is dependent on diffusion, to be reversible traps. Of these, grain boundaries can either

* Corresponding author. Tel.: þ61 (0) 2 80608774.


E-mail address: ayeshajhaq@gmail.com (A.J. Haq).
0360-3199/$ e see front matter Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijhydene.2012.11.127
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6 2545

increase the diffusion of hydrogen by providing faster paths Further, the change in diffusion coefficient with temperature
for diffusion [4e6] or reduce the mobility of hydrogen by was similar for the three materials.
acting as reversible hydrogen trapping sites at nodes and Since pipeline steels are produced by thermo-mechanically
junction points [7e10]. As a result of these two contradictory controlled processes (TMCP), a range of microstructures can-
effects, Ichimura et al. [11] suggested that the hydrogen result depending on the actual processing conditions.
diffusion coefficient will be a maximum at an optimum grain Consequently, the microstructure, grain size, volume fraction
size and Yazdipour et al. [12] confirmed this by modelling as of inclusions and precipitates, and dislocation density can
well as permeation experiments. vary within a particular strength grade. Moreover, segregation
Irreversible traps, on the other hand are those that trap of various elements can occur (depending on the partitioning
hydrogen permanently at temperatures close to ambient and tendency of each element) and inclusions can be concentrated
are associated with high binding energies. These traps include in the slab centre, which solidifies last. Therefore, alloying
non-metallic inclusions and precipitates. Several types of elements and inclusions tend to be concentrated in the cen-
inclusions have been identified in pipeline steels such as treline region of transfer bar (TB) and in the hot rolled strip. It
Al2O3, complex (Fe, Mn)S or double oxide FeO.Al2O3 inclusions, also follows that the chemical composition and microstruc-
mixed compounds containing AleMgeCaeO, Si-ferric carbide ture of the edge differs from that in the centreline region.
and MnS [13,14]. Lee and Lee [15] considered MnS inclusions to Moreover, the difference in the cooling rates of the edge and
be strong irreversible trapping sites for hydrogen. However, centreline can also introduce differences in grain size
Garet et al. [16] reported that MnS particles act as reversible and dislocation density. To minimise the effects of segrega-
trapping sites since they have moderate binding energies. tion, the trend in recent years has been to shift to lower Mn
They also reported that the trapping efficiency increased with levels [36].
sulphur content. In addition to inclusions, precipitates of Ti, Therefore, the aim of this study was to investigate the
Nb and V such as TiC, TiN, NbC, VC, complex Ti,Nb(C,N), effect of microstructure and composition on hydrogen
Ti4C2S2 etc., are all considered to be irreversible traps permeation in X70 pipeline steels. Permeation studies were
[2,17e24]. performed on two pipeline steels having different composi-
In the past few decades, several investigations have dealt tions, namely standard X70 (1.2% Mn) and medium X70 (0.5%
with the influence of various parameters on diffusion in Mn). Samples were investigated from the transfer bar and hot
pipeline steels [14,25e35]. Table 1 summarises diffusion rolled strip stages of production, and from different regions
coefficients and other parameters obtained, for pipeline (centreline and edge). The diffusion coefficients for the
steels, by different authors along with other related informa- different samples were estimated and the observed perme-
tion such as the solutions employed, exit side coating and ation transients were correlated with the microstructural
method used for calculating the diffusion coefficient. features present in the steel specimens.
Andenna and Torella [25] studied the effect of stress on the
permeability of X65 pipeline steels and reported that there
was no effect for the base material with a ferriteepearlite 2. Experimental
microstructure, whereas the hydrogen diffusivity and flux for
quenched material comprising a bainitic microstructure with 2.1. Materials and processing
martensiteeaustenite islands decreases with increasing level
of stress. Park et al. [31] investigated the effect of micro- Standard X70 and medium Mn X70 grade pipeline steels were
structure on the trapping efficiency of X65 pipeline steel and used in this investigation. The chemical compositions of the
reported that the efficiency of trapping increases in the order: steels are listed in Table 2. Samples of size 40 mm  40 mm
degenerated pearlite, bainite and acicular ferrite. The diffu- and thickness slightly greater than 1 mm were obtained by
sivity of hydrogen and density of traps in X70 and X100 steels wire cutting parallel to the rolling plane of the material as
were estimated by Dong et al. [27,28]; and for X80 by Xue et al. shown in Fig. 1. Samples from the edge and centreline regions
[14]. Kim et al. [29] investigated the effect of environmental were taken from the TB of standard X70 steel and hot rolled
factors such as H2S partial pressures and pH values on the strips of both standard and medium Mn X70 grades produced
permeation of hydrogen in high strength low alloy (HSLA) by BlueScope Steel Ltd. To eliminate the effect of residual
steels and found that reversible trapping was affected by H2S strain and to obtain a normalised structure with equiaxed
rather than pH. The effect of precipitates was investigated by grains, samples from the edge region of TB were normalised in
Koh et al. [30] who reported that diffusion of hydrogen was an argon atmosphere at 950  C for 20 min. The notation used
related to the volume fraction of carbonitride precipitates for the samples investigated is given in Table 3.
rather than the ferrite microstructure. Huang et al. [13]
studied the ability of microstructures and inclusions to trap 2.2. Microstructural investigation
hydrogen in three different X120 steels having similar
amounts of Ti,Nb(C,N) precipitates and found that irreversible After the completion of permeability tests, the cross-sections
hydrogen trapping efficiency increases with the volume frac- of the samples were cut from the region exposed to the elec-
tion of non-metallic inclusions such as Al2O3 and MnS. trolyte and prepared for optical metallography using the
Skjellerudsveen et al. [35] studied the diffusion of hydrogen in standard procedures. The samples were etched with 2% Nital
X70 pipeline steel, simulated coarse grain heat affected zone solution and the microstructures were characterised both by
(CGHAZ) and weld metal and found that the diffusion coeffi- optical microscopy and scanning electron microscopy (SEM)
cient decreased in the order: weld metal, X70 steel and HAZ. using a JEOL JSM 6490LV SEM at an excitation voltage of 20 kV.
2546
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6
Table 1 e Summary of permeation parameters for pipeline steel in literature.
Publication Steel Solution Coating Microstructure Method D (10L10 m2sL1) JN (mol H mL2sL1  10L7) JNL (mol H mL1sL1  10L10) Capp
(mol H mL3)

[27] X70 0.5 M H2S04 + Ni F + AF Time lag 0.263 9.85 7.6 28.8
[28] X100 250 mg/l NaAsO2 Ni F+B Time lag 0.01 2.65 1.4 134
[31] X65 NACE Pd DP/AF/B, % M/A, % Breakthrough 4.05 e 9.4 e 0.8e1.4 13e28
F + AF 8.12 5.73
F + B 9.38 4.45
F + DP 3.75 1.28
F + DP 3.93 0.88
[29] X70 NACE Pd F + DP + AF, F + AF + BF Breakthrough 0.5e4.3 e e e
[34] X52 NACE Pd F + P, AF + P Breakthrough 0.15e0.24 e e 2.6  105e
27.41
[33]a X65 0.1 N NAOH e Q&T Time lag 0.9e0.96 e e e
[25]b X65 NACE e F+P Breakthrough 2.4 e e 5.7
B + M/A Breakthrough 1.1 20.1
F+P Time lag 2.5 e e 5.5
B + M/A Time lag 1.2 15.6
[30] API NACE Pd F + P, F + B, F + AF Breakthrough 3.5-9.5 e e e
[14] X80 0.5 M H2S04 Pd F + BF + M/A Time lag 0.2 e 5.2 26
[35] X70 0.1 N NAOH Pd F+P+M Time lag 0.73e0.79 e e e

F - Ferrite, AF e Acicular Ferrite, B e Bainite, DP e Degenerated Pearlite, M/A e Martensite/Austenite, P e Pearlite, BF e Bainitic ferrite, GB e Granular Bainite, Q&T e Quenched and tempered .
a For 1mm thick samples.
b For Zero stress.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6 2547

Table 2 e Chemical compositions of the steels investigated (wt. %).


Steel C P Mn Si Ni Cr Mo Cu Al Nb Ti V S Ca N

X70 0.074 0.012 1.14 0.22 0.024 0.029 0.1 0.023 0.019 0.06 0.02 0.002 0.002 0.0002 0.0041
MX70 0.085 0.01 0.5 0.19 0.018 0.26 0.11 0.011 0.04 0.059 0.035 0.003 0.002 0.001 0.0054

measured with a thermometer. Both surfaces of the specimen


were abraded with 500 grit silicon carbide paper and cleaned
with methanol. After preparation the final thickness of the
sample was measured and recorded (the average of ten
measurements was calculated). No Pd coating was applied at
the exit side of the samples. The cleaned sample was placed in
between the two cells and sealed by tightening the bolts. A
circular area of 2.84 cm2 was exposed to the electrolyte. A
platinum electrode was placed in the anodic cell and a stain-
less steel electrode in the cathodic cell and the steel specimen
formed the working electrode in both cells. These electrodes
Fig. 1 e Schematic showing the location of edge and were then connected to the terminals of the Hyperm unit. The
centreline samples of the strip/transfer bar. anodic side or the detection cell was filled with 0.1 N NaOH
solution, and polarisation was started. A background current
was observed in the amperometer. With time the background
current decreased. When this current decreased to a value
Since the transport of hydrogen was in the normal direction, below 3 mA, it was compensated by means of the zero shift
the grain size, defined in terms of the mean free path in ferrite using a null potentiostat. In effect, the background current
in the normal direction, was measured by the linear intercept was set to zero. Then the cathodic cell was filled with a solu-
technique. The hardness of the samples was determined tion containing 0.1 N NaOH and a promoter, Na2S$9H2O in
using an INDENTEC Vickers hardness machine at a load of order to accelerate the entry of hydrogen into the steel spec-
10 kg. imen. Alkaline solution was used for the permeation tests in
order to reduce the interaction of sulphide inclusions with the
2.3. Electrochemical hydrogen permeation test electrolyte at the steel surface [38]. Initially, concentrations of
1, 10, and 100 g/L of Na2S$9H2O were used to study the effect of
Hydrogen permeability measurements were performed by the promoter concentration. The data indicated that an increase
electrochemical technique developed by Devanathan and in the concentration of the promoter increased the steady
Stachurski [37]. A modified dual cell made of two poly- state current density as well as the permeation rate, i.e., the
carbonate cells was employed. The specimen was held tightly time taken to reach steady state permeation was reduced. The
between the two cells by using rubber gaskets in such a way steady state permeation current observed for 1 g/L of
that the entire system remained leak-tight and the solutions Na2S$9H2O was very low, close to the lower range of the
on either side of the specimen did not mix. The side of the measuring instrument. On the other hand, a higher concen-
membrane, at which hydrogen was to be generated, was tration of promoter, i.e., 100 g/L, led to significant specimen
maintained at a cathodic potential, whereas an anodic corrosion. Therefore 10 g/L of promoter was selected for all the
potential was applied on the detection side of the membrane. experiments. The solutions were prepared from analar grade
The equipment used for applying the required potentials was chemicals and milli Q water. Hydrogen entry into the steel
the Hyperm Mod HM 403. was then facilitated by applying a constant cathodic current of
The permeability tests were carried out at room tempera- 10 mA which gave a current density of 3.52 mA/cm2. Hydrogen
ture. The room temperature was in the range 21e23  C, as permeated through the specimen to the anodic side where
it was instantaneously oxidised giving rise to an equivalent
current in the exit cell which is a direct measure of the out-
put flux of hydrogen. The Hyperm unit was connected to
Table 3 e Sample notation, mean free path in ferrite and a computer through a data logger to record continuous current
hardness. measurements throughout the experiment with data logging
Sample Notation Mean free Average every 10 s.
path (mm) hardness (HV) After the application of the cathodic current on the entry
X70 Edge SME 2.97  0.12 198  7 side, the permeation current was initially zero for a period of
X70 Centreline SMC 4.52  0.18 187  8 time, after which it increased till it reached a steady state
MX70 Edge MME 3.96  0.37 187  6 value. This is referred to as the rise transient. After reaching
MX70 Centreline MMC 4.05  0.38 186  3 steady state, the entry side charging current was interrupted
X70 TB Edge TBE 4.26  0.33 170  6
and the sample was allowed to outgas completely. The current
X70 TB Centreline TBC 5.28  0.34 189  3
Normalised TB NTB 8.93  0.89 142  9
then started decreasing and reached a low value after a period
of time, giving rise to the decay transient. A second transient
2548 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6

was then obtained by starting the current in the entry side where NT is the number of hydrogen trapping sites per unit
again. Triplicate tests were conducted for each sample. volume (m3), NL is the density of the interstitial sites in the
steel and Eb the hydrogen trap binding energy. The values of
2.4. Analysis of permeation transients NL, and Eb were taken as 7.52  1028 m3 and 0.3 eV respec-
tively [28]. The trap density in this study was calculated using
The permeation transients were expressed in normalised the diffusion coefficient calculated from the lag time of the
units i.e., in terms of normalised flux and dimensionless time first transient.
[19,39]. The flux was normalised with respect to the steady
state value and the dimensionless time was obtained by using
the equation, 3. Results

s ¼ DL t=L2 (1) 3.1. Microstructural characterisation


where, DL is the coefficient of diffusion of hydrogen in pure
ferrite and was taken as 7.2  109 m2/s at room temperature Figs. 2 and 3 show typical microstructures of the different
[39,40], t is the time in seconds and L is the thickness of the samples investigated. The samples taken from the edge and
specimen in metres. centreline regions of the standard and medium Mn strips
As mentioned earlier, the permeation current density basically revealed slightly elongated grains of ferriteepearlite.
observed at the exit side is a direct measure of the output flux For the standard Mn alloy, the edge microstructure (SME) and
of hydrogen. From the steady state permeation current the centreline microstructure (SMC) are shown in Fig. 2(a)
density, iNp the hydrogen permeation flux or permeability, JN,
and(b); and those for the medium Mn alloy MME and MMC are
is calculated using the equation: shown in Fig. 2(c) and (d). The optical image taken from the
centreline region of the standard Mn strip (Fig. 2(b)) indicates
JN L ¼ iN
p =ZF (2) that the ferrite grains are more uniform and generally coarser
than those in the edge region (Fig. 2(a)), which is consistent
where, Z is the number of electrons transferred and F is the
with the slower cooling rate experienced by the centreline
Faraday constant (96,487 C mol1). The diffusion coefficients
region during the hot rolling process, compared to the edge. A
can be calculated from the permeation transients using
closer examination of the microstructures indicated that the
a variety of methods as given below.
grain sizes of edge and centreline regions of the standard Mn
Using the breakthrough time tb:
strip were characterised by a bimodal distribution. The edge
Dtb ¼ L2 =15:3tb (3) region exhibited a larger volume fraction of fine grains with
some larger grains which were approximately 5 times the
and in terms of time lag tl:
average size of the finer grains. In contrast, the centreline
DT ¼ L2 =6tl (4) region consisted of a smaller volume fraction of fine grains
and a larger fraction of coarse grains that were about 3 times
where, Dtb is the diffusion coefficient calculated from break-
the average size of the fine grains. Furthermore, the centreline
through time and DT is the diffusion coefficient calculated
samples also exhibited regions with a banded ferriteepearlite
from time lag. The breakthrough time is obtained from the
microstructure. The difference in average grain size is also
intersection of the tangent at the inflection point of the
evident from Table 3 which lists the grain size in terms of
permeation rateetime curve with the initial permeation level
mean free path in the ferrite phase as well as the hardness of
and the time lag is obtained from the time taken for perme-
the various samples. The edge (Fig. 2(c)) and centreline regions
ation rate to reach 0.63 times the steady state value.
(Fig. 2(d)) of the medium Mn strip, on the other hand, exhibited
The diffusion coefficient can also be calculated from the
a more uniform microstructure with very similar mean grain
rising transient using a plot of lnððip  iN N
p Þ=ip Þ against t where,
sizes. This observation is quantified in Table 3. Moreover, no
ip is the measured permeation current at time t and iN p is the
banding of the ferrite and pearlite was evident in the centre-
steady state permeation rate. This plot has a slope of 1/t0,
line region of the medium Mn strip.
where, t0 ¼ L2/p2D. Therefore,
The microstructure of the edge region of transfer bar (TBE)
D ¼ L2 =p2 t0 (5) exhibited a banded structure comprising alternate bands of
bainitic ferrite, labelled 1, and polygonal and/or quasi-polyg-
Another means of calculating the diffusion coefficient is
onal ferrite, labelled 2 (Fig. 3). A magnified image of TBE is
based on the decay transient and employs a plot of lnðip =iN p Þ
shown in Fig. 3(b). The centreline region of the transfer bar
against t. Again this plot has a slope of 1/t0 from which D can
(Fig. 3(c)) exhibited a microstructure of coarser ferrite plates
be calculated using equation (5).
and some quasi-polygonal ferrite which had formed from
The apparent hydrogen solubility Capp (mol H/m3) can be
coarse grained austenite at a relatively slow cooling rate. This
determined by,
coarse lath ferrite structure is often described as granular
Capp ¼ JN L=D (6) bainite [41]. The microstructure of the normalised transfer bar
The trap density was calculated using the equation given by edge sample is shown in Fig. 3(d). The microstructure consists
Dong et al. [28] of equiaxed ferriteepearlite grains.
It is also clear from Table 3 that samples taken from the
 
DL edge of the standard Mn strip exhibited a higher hardness
NT ¼ NL  1 eEb =RT (7)
D than the samples taken from the centreline regions, whereas
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6 2549

Fig. 2 e Optical micrographs of the strip samples (a) standard Mn edge; (b) standard Mn centreline; (c) medium Mn edge; and
(d) medium Mn centreline.

Fig. 3 e Optical micrographs of transfer bar samples (a) edge; (b) its magnified image; (c) centreline; and (d) normalised
transfer bar. 1 and 2 indicate regions of different morphologies: bainitic ferrite and polygonal and/or quasi-polygonal ferrite,
respectively.
2550 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6

the medium Mn samples showed uniform hardness values. permeation current was not as rapid (Fig. 4(b)) i.e., the time
These values correlate well with the grain sizes of the taken to reach steady state was longer. Moreover, it did not
samples. The transfer bar gave the unexpected result that the reach the same permeation current as the first steady state
edge region was softer (170  6 HV) than the core (189  3 HV). value, which indicates that the permeation flux during the
The thickness of the transfer bar ensures a slower near- second transient is smaller than that observed in the first
surface cooling rate than for hot rolled strip, thereby transient.
accounting for the lower hardness of TBE. The higher than
expected core hardness could be due to Mn segregation that These results are similar to those of Manolatos et al. [43] for
lowered the transformation temperature and increased a pure iron specimen using 0.1 N NaOH in both cells. They
strengthening by solute Mn [42]. attributed these observations to a passive oxide layer that
forms on the exit side of the specimen in the absence of Pd
3.2. Permeation transients coating when an anodic potential is applied. According to the
authors, the passive layer formed on the exit side acts as
Fig. 4(a) is the complete permeation curve for a typical X70 a barrier to hydrogen evolution and does not allow the
specimen which includes the first transient, the decay and the hydrogen concentration on the exit side to stabilise. This
second transient. In Fig. 4(b) both the first and second tran- prevents the attainment of stationary conditions and conse-
sients are superimposed. The following observations can be quently affects the steady state permeation flux. This trend of
made from these plots: the permeation transient decreasing with time was also
observed by Beck et al. [44], Amiot et al. [45] and Iino [38]. Beck
1. The first transient current increased rapidly, reached et al. [44] attributed it to void nucleation at an aggregate of
a maximum and remained steady for a short duration of dislocations and Amiot et al. [45] attributed it to the hydrogen
time (<10 min for the NTB samples and about 30e40 min for flux loss through the breakage of surface blisters. However,
the other samples). The current then decreased continu- similar to the present results for all samples, Iino [38] did not
ously even when the sample was still cathodically polarised. observe any blisters. Bruzzoni and Garavaglia [46] also showed
2. Although the start of the second transient (the break- that, in the absence of Pd coating on the exit side, an oxide
through time) appeared to be shorter, the rise in layer is formed, which reduces the permeation flux and also
increases the time taken to reach steady state. Moreover, the
entry side corrodes during the decay since no current is being
applied. This alters the conditions at the hydrogen entry
surface and could result in a lower or delayed hydrogen
uptake during the second transient.
Skjellerudsveen et al. [35] did not observe any significant
variation in the permeation curves of the first, second and
third transients for X70 steel, coated with Pd on the exit side,
at room temperature. The values of steady state current
densities were also similar for the three transients. They also
applied a positive potential to both sides during decay,
contrary to the procedure used in the current investigation.
Our results are in agreement with the conclusions of
Manolatos et al. [43] and Bruzzoni and Garavaglia [46] that, in
the absence of Pd coating, the progressive formation of an
oxide layer on the exit side does not allow a steady state
permeation current to be maintained for a prolonged period
during the permeation test, despite the cathodic polarisation
of the sample. Hence the permeation current decreases after
a relatively short steady state period during the first perme-
ation cycle, as shown in Fig. 4(a). As a consequence of the film
on the exit surface, the second permeation cycle shows
a lower peak permeation current, Fig. 4(a) and (b), before the
permeation current starts to fall as the barrier effect of the
film increases.
The steady state permeation, in this study, did last for
a short period of time, unlike the observations by Manolatos
et al. [43], and made it possible to determine normalised
Fig. 4 e Typical permeation curves for samples examined permeation fluxes for carrying out comparisons. Permeation
in this study. (a) Complete plot of two successive transients obtained for different samples investigated were
permeation cycles for normalised transfer bar sample; and similar to those shown in Fig. 4(a). The most reliable
(b) a plot showing the first and second transients for the comparative data come from the first transient curve up to the
medium Mn edge sample with the second transient time range of the steady state permeation current, because
(dashed line) superimposed on the first transient line. the influence of oxide film formation over this time period
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6 2551

remains relatively low. Accordingly, the curves comparing the difference between the edge and centreline regions was
permeation characteristics of the samples tested, shown in insignificant for the medium Mn strip.
Fig. 6(a)e(g), are for the first transient up to the steady state The various estimates of diffusivity (excluding Dt0 decay)
period. were in reasonable agreement for a given sample. Further,
the trend in diffusivity values for the seven samples
3.3. Diffusion coefficients examined was similar for each of the six different calculation
methods. For example, the average diffusivity for SMC was
Table 4 compares the diffusion coefficients calculated for the 2.63  1010 m2/s with a standard deviation of 0.71.
1st and 2nd transients using the measured breakthrough Variation in hardness and diffusion coefficient with mean
times and lag times, and also from the rising first transient free path has been plotted in Fig. 5 for the standard Mn strip
and decay transient for the different samples investigated in edge, strip centreline and normalised transfer bar, which all
this study. There is a spread in the values of the diffusion had similar ferriteepearlite microstructures. A linear corre-
coefficients obtained from the first transient due to the lation is evident between hardness and mean free path and
different calculation methods, as reported also by Turnbull also the diffusivity and mean free path. Consistent with
and Carroll [39] and Charbonnier et al. [47]. The diffusion literature, the hardness increases as the grain size (the mean
coefficients calculated from breakthrough times are the free path in ferrite) decreases. In contrast, the diffusivity
largest, whereas the diffusion coefficients calculated from lag increases as the mean free path increases. These observations
times and rising first transients are quite similar and lower agree with an earlier study [12] which showed that the diffu-
than the values calculated from the breakthrough times. sion coefficient decreases with decreasing grain size for grains
Values calculated from the decay transient are significantly smaller than 45 mm.
lower, consistent with the significantly longer times taken for
the decay of hydrogen. Garet et al. [16] and Scoppio and Barteri 3.4. Trap densities and apparent hydrogen solubility
[33] have also observed lower diffusion coefficients from
the degassing transient and attributed this to hydrogen The computed values of the density of traps, flux, perme-
exchanges between reversible traps in ferritic steels which ability and the subsurface concentration of hydrogen atoms
result in slow degassing when the hydrogen supply progres- are given in Table 5. The calculated trap densities varied from
sively decreases. Scoppio and Barteri [33] therefore suggest 1.3e3.1  1018 m3 for the different samples, being lowest for
that the phenomena occurring during degassing are complex NTB samples and highest for the MME and MMC samples. This
and hence the parameters estimated from degassing are less trend is consistent with the measured diffusion coefficients
significant compared to the values determined from hydrogen since the trap densities are inversely proportional to the
uptake. diffusion coefficients. Furthermore, there was no difference in
The shorter breakthrough time for the second transient, the trap densities of the edge and centreline regions for the
correlates with the consistently higher diffusion coefficient medium Mn strip.
(1.11e1.79 times higher) than that for the first transient (Table The apparent hydrogen solubility is also inversely related
4). Table 4 also indicates that there are only minor differences to the diffusion coefficient. Therefore, there were only minor
in the diffusion coefficients calculated from the breakthrough differences in the apparent hydrogen solubility of the samples
time for the first transient for the different samples tested. investigated.
The medium Mn edge sample exhibited the lowest diffusivity
(1.88  1010 m2/s) and the normalised transfer bar showed
the highest diffusivity (5.66  1010 m2/s). For the standard Mn 4. Discussion
strip, the samples taken from the edge of the strip gave a lower
diffusion coefficient than the centreline region, whereas the 4.1. Comparison of the estimated permeation
parameters with literature

In addition to the microstructure and composition of the steel,


the diffusion of hydrogen can be influenced by other experi-
mental factors including the solution used for permeability
testing, the concentration of promoter (if any), the method
used to calculate the diffusion coefficient and the presence or
absence of Pd or an other coating on the exit side of the
sample. Serna et al. [34] have also pointed out that the diffu-
sion coefficient can depend on the charging conditions.
Therefore, although several researchers have studied the
diffusion of hydrogen in micro-alloyed pipeline steels, it is not
easy to compare the diffusion coefficients. This is also evident
from Table 1 which shows that the diffusion coefficients ob-
tained by different researchers vary significantly.
Fig. 5 e Variation in hardness and diffusion coefficient Tables 1 and 4 show that the values of diffusion coeffi-
with mean free path for samples having similar cient obtained in this study are an order of magnitude higher
ferriteepearlite structure and composition. than those reported by Dong et al. [27] and Skjellerudsveen
2552 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6

Fig. 6 e Comparison of the first permeation transients for (a) standard Mn strip edge and centreline samples; (b) medium Mn
strip edge and centreline; (c) standard and medium Mn strip edge; (d) standard and medium Mn strip centreline; (e) transfer
bar edge and centreline; (f) transfer bar edge and normalised transfer bar edge; and (g) normalised transfer bar and hot rolled
strip of standard Mn steel.

et al. [35] for X70 steel. The former researchers used an acidic Dong et al. [27]. The reported microstructures were also
solution of H2SO4 with 250 mg/L As2O3 and the latter 0.1 N different.
NaOH without any promoter. Furthermore, the exit side On the other hand, the results obtained from the present
was coated with Pd by Skjellerudsveen et al. [35] and Ni by study are comparable to those reported by Park et al. [31],
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6 2553

Table 4 e Diffusion coefficients calculated by different methods.


Sample Dtb (1010 m2 s1) DT (1010 m2 s1) Dt0 rise (1010 m2 s1) Dt0 decay (1011 m2 s1)

First Second First Second

SME 2.29  0.05 4.09  1.5 1.97  0.09 1.77  0.06 2.0  0.09 5.90  0.2
SMC 3.02  0.07 3.71  0.7 2.25  0.19 2.06  0.17 2.14  0.15 5.62  0.8
MME 1.88  0.19 3.00  0.44 1.70  0.04 1.57  0.03 1.67  0.07 3.88  0.5
MMC 2.10  0.21 2.54  0.61 1.71  0.1 1.52  0.11 1.59  0.14 4.82  0.5
TBE 3.70  0.02 4.58  0.01 2.10  0.19 1.33  0.31 1.68  0.4 3.84  0.8
TBC 4.87  0.74 5.4  0.63 3.05  0.17 2.15  0.30 2.8  0.19 6.04  0.7
NTB 5.66  0.23 7.46  0.45 4.01  0.02 3.82  0.13 3.62  0.58 7.87  1.1

despite their use of NACE solution for the entry side and 4.2.1. Comparison of standard Mn strip edge and centreline
Pd coating on the exit side. Diffusion coefficients of Fig. 6(a) shows the first permeation transient of the SME and
2.3e3  1010 m2/s were calculated, in the present study, from SMC samples. It is apparent from this figure that not only is
breakthrough times of the first transient for the standard X70 the breakthrough time longer in the edge samples, but the
strip comprising a ferriteepearlite structure. These results are transient is less steep, which implies that it contains a higher
about 3 times lower than the values of 9.27e9.38  1010 m2/s density of traps which reduce the mobility of hydrogen.
obtained by Park et al. [31] for ferrite-degenerated pearlite Therefore it takes a longer time to fill the traps and conse-
structures in X65 steel. However, a diffusion coefficient of quently reach a steady state permeation current
3.7  1010 m2/s obtained for the ferriteebainitic ferrite During continuous casting, Mn has a strong tendency to
structure (TBE) in this study is close to the value of segregate to the centreline region of the slab. Escobar et al. [48]
4.05  1010 m2/s reported by Park et al. [31] for their A2 steel have demonstrated by etching with LePera reagent that Mn
with a similar microstructure. The present results are also segregation occurs in hot rolled steels. As a result of this
comparable to the values reported by Koh et al. [30] for pipe- segregation, Mn can combine with S during the last stages of
line steels having ferriteepearlite microstructures. solidification giving rise to MnS inclusions. These inclusions
Although the diffusion coefficients are of the same order have been reported to act as moderate to strong trapping sites
of magnitude, the values of permeability and subsurface which can trap hydrogen reversibly [16] or irreversibly [15] and
concentration of hydrogen observed in this study are thereby reduce the diffusion of hydrogen. Consequently, the
significantly lower that those reported by Park et al. [31]. This centreline samples should exhibit a lower diffusion coeffi-
is possibly related to the different solutions used for the cient. However, the samples that were used in the permeation
entry side as well as the presence of Pd coating in the case of studies did not appear to show any pronounced segregation in
Park et al. [31] which allows full oxidation of the hydrogen terms of the presence of MnS inclusions. In view of the above,
arriving at the exit side and gives rise to a higher permeation the results obtained here cannot be explained on the basis of
flux. concentration of MnS particles; rather they can be rationalised
on the basis of the microstructure of the samples. As seen in
Fig. 2(a) and (b) and Table 3, both the edge and centreline
4.2. Comparisons of the different samples studied samples of the standard Mn strip show non-uniform micro-
structures with a bimodal distribution of grains. However, the
As discussed earlier, the diffusion coefficients of the different edge sample has a larger number of very fine grains and
samples tested here were of the same order of magnitude, whereas the fraction of grains having a larger size is higher in
with only some minor differences. However, if the first the centreline samples, resulting in a lower mean free path for
permeation transients of the different sets of sample condi- the edge sample as compared to the centreline sample. As
tions are plotted on the same graph, as shown in Fig. 6(a)e(g), shown in an earlier publication [12] for X70 steel, ferrite grain
clear differences in the permeation behaviour of the samples sizes smaller than 45 mm can reduce the mobility of hydrogen
are more noticeable. These differences are discussed below. by trapping at nodes and triple junctions; with the extent of

Table 5 e Trap density, flux, permeability and solubility.


Sample NT (1018 m3) Flux JN (107 mol H m2 s1) Permeability JNL (1010 mol H m1 s1) Capp (101 mol H m3)

SME 2.68  0.12 1.54  0.26 1.66  0.25 8.40  0.93


SMC 2.37  0.21 1.92  0.22 1.65  0.23 7.38  0.56
MME 3.12  0.08 1.82  0.26 1.95  0.25 11.5  1.5
MMC 3.10  0.18 1.58  0.30 1.55  0.35 9.02  1.5
TBE 2.51  0.23 1.60  0.06 1.67  0.16 8.02  1.5
TBC 1.70  0.1 2.22  0.26 2.34  0.20 7.69  0.91
NTB 1.28  0.01 3.43  0.26 2.42  0.03 6.03  0.11
2554 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6

this trapping increasing with decreasing grain size. Conse- C and N contents of the medium Mn X70 steel were reported
quently, the large number of finer grains in the edge region to result in a significantly larger fraction of carbonitride
could increase the trapping of hydrogen and thereby give rise precipitates. Fine precipitates of TiC, NbC or complex car-
to a lower diffusion coefficient. In addition, it can be expected bonitrides of Ti and Nb have been found to act as strong
that the edge region of the strip would contain a higher irreversible traps [2,18,19]. Permeability studies in pipeline
density of dislocations as compared to the centreline region, steels by Koh et al. [30] have also shown that diffusivity of
because the faster cooling rate during processing of the strip hydrogen decreases as the density of precipitates increases.
promotes transformation to ferrite at lower temperature with Therefore, the lower diffusivity of the medium Mn strip
an increase in dislocation density, arising from the trans- compared to the standard Mn strip can be attributed to the
formation volume change. Since dislocations act as reversible presence of these precipitates which trap hydrogen effec-
traps for hydrogen, the combined effect of a lower grain size tively and reduce its mobility. This is also consistent with the
and higher dislocation density of the edge region of the strip higher density of traps calculated for the medium Mn strip, as
could result in the lower diffusivity of hydrogen seen in Table 5
Some banding of ferriteepearlite grains parallel to the The permeation transients of the centreline regions of the
rolling direction of the strip was also observed in the centre- two steels also show a similar trend (Fig. 6(d)) with a slightly
line samples. Since the samples for permeability tests were larger difference in the two strips compared to the edge
obtained by cutting slices parallel to the rolling plane, this region. Again the lower diffusivity of the medium Mn steel can
means that the banding was perpendicular to the direction of be attributed to the presence of precipitates. In addition, the
hydrogen entry and movement. The effective diffusivities of mobility of hydrogen in the MMC samples is further reduced
hydrogen in such specimens have been reported to be an by the contribution to trapping because of a slightly lower
order of magnitude lower than when hydrogen entry was mean free path compared to the SMC samples (Table 3).
along the direction of banding [49,50]. However, in this study Therefore, the permeation transients show a slightly greater
the small amount of banding did not appear to have any difference, relative to those for the edge samples
significant effect in decreasing the diffusion of hydrogen in
the centreline samples as compared to the non-banded edge 4.2.4. Comparison of standard Mn transfer bar edge and
samples centreline
Fig. 6(e) shows the permeation plots for the edge and centre-
4.2.2. Comparison of medium Mn strip edge and centreline line regions of the X70 TB. This figure indicates that diffusion
Comparison of the permeation transients of the two regions of is faster in the centreline region of the TB than the edge
the medium Mn strip (Fig. 6(b)) indicates that there is little region. This effect can be attributed to the coarse granular
difference in the form of the transients and therefore their bainite microstructure formed from coarser prior austenite
hydrogen diffusivity values are similar (Table 4). Examination grains of TB centreline samples. In contrast, the edge region
of the microstructures of the two regions revealed that the has a mixture of finer bainitic ferrite laths and quasi-
grain sizes were uniform and comparable, with the edge polygonal ferrite. In addition to the bainitic ferrite grains
region showing only a marginally lower mean free path. being fine with a lower mean free path, the edge microstruc-
Therefore, the similarity of the mean free paths in the ferrite ture is expected to have a high density of dislocations which
phase for the two regions is consistent with the similarity of can also trap hydrogen and reduce its mobility. Therefore, the
the permeation transients. TBE samples exhibit a lower diffusivity compared to the TBC
samples.
4.2.3. Comparison of standard Mn and medium Mn strip
The plot comparing the edge regions of the standard and 4.2.5. Comparison of standard Mn TB edge and normalised
medium Mn strips is shown in Fig. 6(c). The breakthrough time TB edge
and lag time for the medium Mn edge sample were longer The transient for the NTB sample is compared with the TBE
than for the standard Mn edge sample, indicating that trap- sample in Fig. 6(f). Not only is the breakthrough time for the
ping is more effective in the medium Mn strip. An examina- NTB shorter, the transient is also significantly steeper than the
tion of the microstructure of the edge regions of the two steels TBE sample, so that the steady state is attained in a shorter
reveals that the standard Mn strip has a larger number of time. These observations suggest that the diffusion of
grains having a finer grain size and consequently a smaller hydrogen is very fast through the NTB microstructure due to
mean free path in the ferrite phase. As discussed earlier, the absence of features that can act as effective traps for
a finer grain size is expected to trap hydrogen and lower its hydrogen. This is clearly a consequence of the large polygonal
mobility. In view of the grain size effect, the SME samples ferrite grains of the normalised microstructure, which is re-
should have shown a lower diffusivity, i.e., longer break- flected in its significantly larger mean free path (Table 3). In
through and lag times. However, the permeation results show addition, the dislocation density is low due to the normalising
the opposite treatment so that the trapping potential is minimised for this
Recent studies involving precipitate analysis of the two sample. In contrast, the TBE sample had fine bainitic ferrite
steels using scanning electron microscopy [51] have shown structure, smaller grains of quasi-polygonal ferrite and
that the medium Mn steel exhibits a significantly larger a higher dislocation density. These factors combine to effec-
number density of precipitates and inclusions. Most of these tively trap hydrogen and reduce its mobility, giving rise to
precipitates were found to be complex carbonitrides of Ti and a longer breakthrough time and a less steep permeation
Nb having an average diameter of 0.8  0.8 mm. The higher Ti, transient.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6 2555

4.2.6. Comparison of normalised TB and hot rolled strip of the thermal analysis technique. Surface and Coatings
standard Mn steel Technology 1986;28:301e14.
[2] Pressouyre GM, Bernstein I. A quantitative analysis of
Fig. 6(g) shows a comparison of the first transients obtained
hydrogen trapping. Metallurgical Transactions A 1978;10:
for the NTB and SME samples, both of which have a ferri-
1571e80.
teepearlite microstructure. The normalised microstructure [3] Villalba E, Atrens A. SCC of commercial steels exposed to
consists of coarse equiaxed grains with a relatively low high hydrogen fugacity. Engineering Failure Analysis 2008;
dislocation density whereas the hot rolled strip has fine and 15:617e41.
somewhat elongated grains that are expected to have a higher [4] Calder RD, Elleman TS, Verghese K. Grain boundary diffusion
dislocation density. Accordingly, the hardness value of of tritium in 304- and 316-stainless steels. Journal of Nuclear
Materials 1973;46:46e52.
198  7 HV for the hot rolled strip (Table 3) is significantly
[5] Kimura A, Birnbaum HK. Hydrogen induced grain boundary
higher than for the normalised transfer bar (142  9 HV).
fracture in high purity nickel and its alloys-enhanced
Relative to the NTB sample, there was a substantial increase in hydrogen diffusion along grain boundaries. Acta
the breakthrough time and the time to attain a steady state for Metallurgica 1988;36:757e66.
the SME sample, because the finer grains and the higher [6] Tsuru T, Latanision RM. Grain boundary transport of
density of dislocations effectively trap hydrogen and signifi- hydrogen in nickel. Scripta Metallurgica 1982;16:575e8.
cantly reduce its mobility, compared to the coarse equiaxed [7] Brass AM, Chanfreau A. Electrochemical permeation of
hydrogen in high purity nickel at 100  C. Scripta Metallurgica
grains in the normalised microstructure.
et Materialia 1990;24:499e504.
[8] Choo WY, Lee JY, Cho CG. Effect of grain size. Materials
Science 1981;16:1285e93.
5. Conclusions [9] Wert CA. Trapping of hydrogen in metals. In: Alefeld G,
Volkl J, editors. Hydrogen in metals II. Application-oriented
properties. Springer-Verlag; 1978. p. 305e30.
➢ The medium Mn strip showed the lowest diffusivity due
[10] Yao J, Cahoon JR. Theoretical modeling of grain boundary
to the combined effect of a fine grain size and a high
diffusion of hydrogen and its effect on permeation curves.
density of (Ti,Nb)C,N precipitates. Acta Metallurgica et Materialia 1991;39:119e26.
➢ The normalised transfer bar showed the highest diffu- [11] Ichimura M, Sasajima Y, Imabayashi M. Grain boundary
sivity due to its large ferrite grains and a very low effect on diffusion of hydrogen in pure aluminum. Materials
dislocation density. Transactions JIM 1991;32:1109e14.
➢ Similar grain sizes for the edge and centreline samples [12] Yazdipour N, Haq AJ, Muzaka K, Pereloma EV. 2D modelling
of the effect of grain size on hydrogen diffusion in X70 steel.
of medium Mn X70 strip resulted in exhibiting similar
Computational Materials Science 2012;56:49e57.
hydrogen diffusivities, whereas standard X70 edge [13] Huang F, Liu J, Deng ZJ, Cheng JH, Lu ZH, Li XG. Effect of
samples showed lower diffusivities than the centreline microstructure and inclusions on hydrogen induced cracking
samples due to the combination of finer grain size and susceptibility and hydrogen trapping efficiency of X120
a higher dislocation density. pipeline steel. Materials Science and Engineering A 2010;527:
➢ The low dislocation density and coarser ferriteepearlite 6997e7001.
[14] Xue HB, Cheng YF. Characterization of inclusions of X80
structure of the normalised TB edge sample exhibited
pipeline steel and its correlation with hydrogen-induced
a significantly higher diffusivity than the SME sample
cracking. Corrosion Science 2011;53:1201e8.
despite the similarity of the microstructure. [15] Lee SM, Lee JY. The effect of the interface character of TiC
➢ The fine structure of bainitic ferrite and associated high particles on hydrogen trapping in steel. Acta Metallurgica
dislocation densities of the TB edge sample resulted in 1987;35:2695e700.
a lower diffusivity than the granular bainite formed from [16] Garet M, Brass AM, Haut C, Guttierez-Solana F. Hydrogen
coarser prior austenite grains of the TB centreline trapping on non metallic inclusions in CreMo low alloy
steels. Corrosion Science 1998;40:1073e86.
sample.
[17] Stevens MF, Bernstein IM. Microstructural trapping effects
on hydrogen induced cracking of a microalloyed steel. Metall
Trans A 1989;20A:909e19.
Acknowledgements [18] Takahashi I, Matsumoto Y, Tanaka T. In: Proc. JIMIS-2.
Minakami, Tokyo, Japan; 1979.
[19] Valentini R, Solina A, Matera S, De Gregorio P. Influence of
This study was conducted as part of Australian Research
titanium and carbon contents on the hydrogen trapping of
Council (ARC) Linkage Grant (LP0883546) with BlueScope Steel microalloyed steels. Metallurgical and Materials
Ltd. (BSL) as the industrial partner. The authors would like to Transactions A 1996;27:3773e80.
thank the ARC and BSL for their financial support. The authors [20] Wei FG, Tsuzaki K. Quantitative analysis on hydrogen
would also like to thank BSL for providing the Hyperm Mod trapping of TiC particles in steel. Metallurgical and Materials
HM 403 unit for carrying out the permeability measurements. Transactions A 2006;37:331e53.
[21] Wei FG, Hara T, Tsuzaki K. Nano-precipitates design with
hydrogen trapping character in high strength steels. In:
Somerday B, Sofronis P, Jones R, editors. Effects of hydrogen
references on materials, proceedings of the international hydrogen
conference, Wyoming, USA; 2008. p. 448e55.
[22] Wei FG, Tsuzaki K. Hydrogen trapping character of nano-
[1] Lee JY, Lee SM. Hydrogen trapping phenomena in metals sized NBC precipitates in tempered martensite. In: Somerday
with B.C.C. and F.C.C. crystals structures by the desorption B, Sofronis P, Jones R, editors. Effects of hydrogen on
2556 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 5 4 4 e2 5 5 6

materials, proceedings of the international hydrogen [36] Williams JG. Advances in steels for high strength ERW
conference, Wyoming, USA; 2008. p. 456e63. linepipe applications in Australia. Materials Forum 2007;31:
[23] Yamasaki S, Takahashi T. Evaluation method of delayed 1e10.
fracture property of high strength steels. Tetsu-to-Hagane [37] Devanathan MAV, Stachurski Z. The adsorption and
1997;83:454e9. diffusion of electrolytic hydrogen in palladium. The Royal
[24] Yuan X. Precipitates and hydrogen permeation behavior in Society of London A 1962;270:90e102.
ultra-low carbon steel. Materials Science and Engineering A [38] Iino M. Analysis of irreversible hydrogen trapping. Acta
2007;452-453:116e20. Metallurgica 1982;30:377e83.
[25] Andenna C, Torella R. A study of hydrogen permeation on [39] Turnbull A, Carroll MW. The effect of temperature and H2S
API X65 TMCP steel under stress conditions. In: Turnbull A, concentration on hydrogen diffusion and trapping in a 13%
editor. Hydrogen transport and cracking in metals. chromium martensitic stainless steel in acidified NaCl.
Proceedings of a conference, Teddington UK; 1995. p. 240e52. Corrosion Science 1990;30:667e79.
[26] Contreras A, Albiter A, Salazar M, Pérez R. Slow strain rate [40] Kiuchi K, McLellan RB. The solubility and diffusivity of
corrosion and fracture characteristics of X-52 and X-70 hydrogen in well-anealed and deformed iron. Acta
pipeline steels. Materials Science and Engineering A 2005; Metallurgica et Materialia 1983;31:961e84.
407:45e52. [41] Dunne DP. Ferrite morphology and residual phases in
[27] Dong CF, Li XG, Liu ZY, Zhang YR. Hydrogen-induced continuously cooled low carbon steels. Materials Forum
cracking and healing behaviour of X70 steel. Journal of Alloys 1999;23:63e76.
and Compounds 2009;484:966e72. [42] Desforges CD, Duckworth WE, Ryan TFJN. Manganese in
[28] Dong CF, Liu ZY, Li XG, Cheng YF. Effects of hydrogen- ferrous metallurgy. Paris: The Manganese Centre; 1977.
charging on the susceptibility of X100 pipeline steel to [43] Manolatos P, Jerome M, Thual CD, Coze JL. The
hydrogen-induced cracking. International Journal of electrochemical permeation of hydrogen in steels without
Hydrogen Energy 2009;34:9879e84. palladium coating. Part I: interpretation difficulties.
[29] Kim WK, Koh SU, Yang BY, Kim KY. Effect of environmental Corrosion Science 1995;37:1773e83.
and metallurgical factors on hydrogen induced cracking of [44] Beck B, Bockris JO, McBreen J, Nanis L. Proceedings of the
HSLA steels. Corrosion Science 2008;50:3336e42. Royal Society of London A 1966;290:220.
[30] Koh SU, Kim JS, Yang BY, Kim KY. Effect of line pipe steel [45] Amiot PAB, Pinard-Legry G. Contribution a I’etude du
microstructure on susceptibility to sulfide stress cracking. cloquage des a&r pour constructions soudees. presentee
Corrosion Science 2004;60:244e54. a Deuxieme Congres International. L’hydrogene dans les
[31] Park GT, Koh SU, Jung HG, Kim KY. Effect of microstructure metaux. Paris, 3E8; 1977.
on the hydrogen trapping efficiency and hydrogen induced [46] Bruzzoni P, Garavaglia R. Anodic iron oxide films and their
cracking of linepipe steel. Corrosion Science 2008;50:1865e71. effect on the hydrogen permeation through steel. Corrosion
[32] Ramı́rez E, González-Rodriguez JG, Torres-Islas A, Serna S, Science 1992;33:1797e807.
Campillo B, Dominguez-Patiño G, et al. Effect of microstructure [47] Charbonnier JC, Margot-Marette H, Brass AM, MAucoutier M.
on the sulphide stress cracking susceptibility of a high strength Sulfide stress cracking of high strength modified
pipeline steel. Corrosion Science 2008;50:3534e41. CreMo steels. Metallurgical Transactions A 1985;16:935e44.
[33] Scoppio L, Barteri M. Methods of hydrogen uptake [48] Pérez Escobar D, Miñambres C, Duprez L, Verbeken K,
measurements by electrochemical permeation test on low Verhaege M. Internal and surface damage of multiphase
alloy steels. In: Turnbull A, editor. Hydrogen transport and steels and pure iron after electrochemical hydrogen
cracking in metals. Proceedings of a conference, Teddington charging. Corrosion Science 2011;53:3166e76.
UK; 1995. p. 204e15. [49] Lee HL, Chan SLI. Hydrogen embrittlement of AISI 4130 steel
[34] Serna S, Martı́nez H, López SY, González-Rodrı́guez JG, with an alternate ferrite/pearlite banded structure. Materials
Albarrán JL. Electrochemical technique applied to Science and Engineering A 1991;142:193e201.
evaluate the hydrogen permeability in microalloyed steels. [50] Tau L, Chan SLI. Effects of ferrite/pearlite alignment on the
International Journal of Hydrogen Energy 2005;30:1333e8. hydrogen permeation in a AISI 4130 steel. Materials Letters
[35] Skjellerudsveen M, Akselsen OM, Olden V, Johnsen R, 1996;29:143e7.
Smirnova A. Effect of microstructure and temperature on [51] Hejazi D, Haq AJ, Yazdipour N, Dunne DP, Calka A, Barbaro F,
hydrogen diffusion and tripping in X70 grade pipeline steel et al. Effect of manganese content and microstructure on the
and its weldments. In: Proceedings of the 2010 EUROCORR susceptibility of pipeline steel to hydrogen cracking.
conference, Moscow, Russia; 2010. Materials Science and Engineering A 2012;551:40e9.

You might also like